Re: [squid-users] Slow to Stop - 2.5-STABLE on Solaris 8

From: Jesse Reynolds <[email protected]>
Date: Wed, 26 Nov 2003 22:12:17 +1100

At 11:58 +0100 26/11/03, Henrik Nordstrom wrote:
>On Wed, 26 Nov 2003, Jesse Reynolds wrote:
>
>> Most of the time squid is taking about 30 seconds to shut down for
>> me. I have built 2.5-STABLE on Solaris 8. The same thing happens on
>> all 7 boxes here.
>>
>> Occasionally it stops very quickly, but usally it just takes ages to stop.
>
>What is said in cache.log?
>
>Maybe your squid is simply configured to wait up to 30 seconds for clients
>to finish their requests (this is the default).

Ah right, thankyou Henrik!

It's odd that it seems to do this most of the time, even when noone
is accessing the cache. Perhaps it is also considering the file
descriptors for the child redirector processes?

 From the squid.conf:

# TAG: shutdown_lifetime time-units
# When SIGTERM or SIGHUP is received, the cache is put into
# "shutdown pending" mode until all active sockets are closed.
# This value is the lifetime to set for all open descriptors
# during shutdown mode. Any active clients after this many
# seconds will receive a 'timeout' message.
#
#Default:
# shutdown_lifetime 30 seconds

Perhaps I'll set this to 5 seconds. ... There are four squids in
accelerator mode behind a load balancer. We want to be able to take
one out of service with minimal impact on user experience. Currently
you randomly get a "squid is shutting down" error if you do this.
Closer interaction with the Cisco CSM load balancer could also be a
better way to do this!

Cheers

Jesse

-- 
   ::: Jesse Reynolds +61 (0)414 669 790 ::: AIM - jessedreynolds :::
   ::: Virtual Artists Pty Ltd, Adelaide ::: http://www.va.com.au :::
Received on Wed Nov 26 2003 - 04:12:28 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:21:37 MST